Tar: Not able to write a PaxTarEntry from a TarReader.

[jozkee]

Consider this method that creates a new archive copying the entries from an existing one:

static void CopyTar(string archive, string destination)
{
    Directory.CreateDirectory(destination);
    using var reader = new TarReader(File.OpenRead(archive));
    using var writer = new TarWriter(File.Create(Path.Combine(destination, "copy_tar")));

    TarEntry? e;
    while ((e = reader.GetNextEntry()) != null)
    {
        writer.WriteEntry(e);
    }
}

When the archive is in pax format, it throws the following exception:

Unhandled exception. System.ArgumentException: An item with the same key has already been added. Key: path
   at System.Collections.Generic.Dictionary`2.TryInsert(TKey key, TValue value, InsertionBehavior behavior)
   at System.Collections.Generic.Dictionary`2.Add(TKey key, TValue value)
   at System.Formats.Tar.TarHeader.CollectExtendedAttributesFromStandardFieldsIfNeeded()
   at System.Formats.Tar.TarHeader.WriteAsPax(Stream archiveStream, Span`1 buffer)
   at System.Formats.Tar.TarWriter.WriteEntry(TarEntry entry)
   at Program.<<Main>$>g__Extract|0_0(String archive, String destination, TarEntryFormat format) in /Users/jozkee/dotnet-consoles/tar_utf8/Program.cs:line 17
   at Program.<Main>$(String[] args) in /Users/jozkee/dotnet-consoles/tar_utf8/Program.cs:line 5

This is because the TarWriter attempts to add "Path" to the Extended Attributes when the entry read with TarReader already contained one.

This is somewhat important since PAX is our default format, but there may be a workaround, that is, use the new PaxTarEntry(TarEntryType entryType, string entryName, IEnumerable<KeyValuePair<string, string>> extendedAttributes) ctor.

Tagging subscribers to this area: @dotnet/area-system-io
See info in area-owners.md if you want to be subscribed.

Issue Details

---

Consider this method that creates a new archive copying the entries from an existing one:

static void CopyTar(string archive, string destination)
{
    Directory.CreateDirectory(destination);
    using var reader = new TarReader(File.OpenRead(archive));
    using var writer = new TarWriter(File.Create(Path.Combine(destination, "copy_tar")));

    TarEntry? e;
    while ((e = reader.GetNextEntry()) != null)
    {
        writer.WriteEntry(e);
    }
}

When the archive is in pax format, it throws the following exception:

Unhandled exception. System.ArgumentException: An item with the same key has already been added. Key: path
   at System.Collections.Generic.Dictionary`2.TryInsert(TKey key, TValue value, InsertionBehavior behavior)
   at System.Collections.Generic.Dictionary`2.Add(TKey key, TValue value)
   at System.Formats.Tar.TarHeader.CollectExtendedAttributesFromStandardFieldsIfNeeded()
   at System.Formats.Tar.TarHeader.WriteAsPax(Stream archiveStream, Span`1 buffer)
   at System.Formats.Tar.TarWriter.WriteEntry(TarEntry entry)
   at Program.<<Main>$>g__Extract|0_0(String archive, String destination, TarEntryFormat format) in /Users/jozkee/dotnet-consoles/tar_utf8/Program.cs:line 17
   at Program.<Main>$(String[] args) in /Users/jozkee/dotnet-consoles/tar_utf8/Program.cs:line 5

This is because the TarWriter attempts to add "Path" to the Extended Attributes when the entry read with TarReader already contained one.

This is somewhat important since PAX is our default format, but there may be a workaround, that is, use the new PaxTarEntry(TarEntryType entryType, string entryName, IEnumerable<KeyValuePair<string, string>> extendedAttributes) ctor.

Author:	Jozkee
Assignees:	-
Labels:	area-System.IO
Milestone:	-

[jozkee]

cc @carlossanlop

[stephentoub]

FWIW, I hit this as the very first thing I tried with TarReader/Writer, as part of writing a perf test to roundtrip a tar file. I forgot to open a bug on it and was only just reminded of it when I saw your issue, so thanks for opening this. If there's a straightforward fix, I think we should fix it for 7.